RE: URIs for Physical Items

You wrote: "I assume you meant URL as a synonym for URI.)"

No.  Precisely not.  Actually, this illustrates the point I was trying to
make. I think obfuscating URLs and URIs does not help the discussion. I
simply use URLs as the guy in the street understands that term: what he sees
on the side of a bus.  Because thats all thats out there on the web for all
practical purposes right now.  

The issue I responded to (and all I responded to, I thought) was the
proposition that a UR*L* is appropriate for identifying physical items.  I
stand by what I said in that case. I also have been misread: I fundamentally
AGREE that we need identifiers for classes, for multiple instances, for lots
and lots of things.  And I think URLs don't always do that (e.g. identify
clases; enable identifcation of multiple copies).  And (here goes) I think
URIs may do, -  but right now are too woolly in their definition and not yet
widely practically implemented other than in the case of that subset of URIs
which are URLs... and I know I will be assailed from all sides for saying
that so i will now  retreat to the W3C URI activity which is meant to
address some of these issues.  


-----Original Message-----
From: Dan Connolly [mailto:connolly@w3.org]
Sent: 17 October 2000 06:23
To: Paskin, Norman (DOI-ELS)
Cc: 'uri@w3.org'
Subject: Re: URIs for Physical Items


"Paskin, Norman (DOI-ELS)" wrote:
> 
> No, URLs are not appropriate for designating physical items:
> 1. A physical item may exist in multiple copies.  ISBNs for example do not
> refer to a specific copy of a book; they identify the class of all copies
in
> an edition.  It is useful to identify the class not a specific instance of
> it.

An ISBN is not a suitable identifier if you mean to
distinguish one copy of a book from another, but
it seems perfectly reasonable to refer to
the publication of a book, i.e. the social/legal event
of making >=6 copies of an intellectual work
available to the public, as isbn:XXXXXX.

Further, just because isbn:xxx isn't a suitable
identifier for one copy of a book as distinguished
from another, it does not follow that no URI is
an appropriate identifiers for any book. (hmm... your
conclusion regards URLs; I assume you meant
URL as a synonym for URI.)

As long as you can make the denotation of a URI
indistinguishable from a physical object, 
it's fine to use that URI to identify that object.

For example, http://www.w3.org/ is a poor
identifier for the particular book I'm looking
at on my desk, because it's pretty easy
to see that the title of one is "World Wide
Web Consortium" and the title of the other
is "Extreme Markup Languages 2000".

But consider:
	uuid:c060d0d-103-501-b07-1090c0030e

I hereby allocate that URI to denote the copy of
the XML 2000 proceedings on on my desktop.

The following characteristics are sufficient
to identify it and to distinguish it from
all other physical objects:

	* it is a book, loosely speaking.
	In particular, its a conference proceedings.
	It has 255 numbered pages.

	* it was edited by Steven R. Newcomb
	and B. Tommie Usdin in preparation
	for the XML 2000 conference in Montreal.

	* it (among other things) was given to me
	in Montreal in August 2000
	in exchange for a my conference registration
	payment of $525.00 US;
	invoice for which payment is dated 04 August 2000
	and numbered 0000093096
	from GCA, the holders of the domain www.gca.org
	on that date; and in which invoice GCA offers to answer
	HTTP requests to http://www.gca.org/
	and inquiries sent by internet mail to info@gca.org,
	and telephone calls to +1-703-519-8160 .

Now one problem with that identifier is that uuid:
isn't (yet) registered with IANA; if it were, I would have
an agreement with the Internet community about
the denotation of that name by sending this message:
the IETF process presumably operates by consent of the governed,
since any Internet user may participate in the process
and seek redress of grievances or whatever; that process
allocates uuid:* as per the UUID specs, and I have
a conforming implementation of the UUID specs
on my desktop, and I just used it.

But let's take a look at a few identifiers that I
do have social/legal authority to bind (and
which I do hereby bind):

	http://www.w3.org/2000/10/c060d0d-103-501-b07-1090c0030e

If you try to access that with your web browser, you
may discover, by way of the "403 Forbidden" response
that I do not authorize you to observe
my book via HTTP. I'm pretty sure it's impossible
to refute my claim that this identifier denotes
this book. But using the http URI scheme does
have some practical connotations that make
it awkward to use as a book, and I'm not
certain of the logical indistinguishability.

But I'm pretty certain this one, which,
again, I do hereby bind to denote that book,
cannot be distinguished from the book
by any logical argument:

	http://www.w3.org/2000/10/#c060d0d-103-501-b07-1090c0030e

That is: I'm pretty certain that URI references that include
fragment identifiers can be bound to the same
denotation as *any* natural language definite description[dd]
without any logical problems.


[dd] http://www.w3.org/Architecture/Terms#definite-description

> 2. A user may well wish to differentiate between a website (URL) -e.g. for
> maintenenace, administration; and the entity currently avaiulable at that
> website.  If the URl is used for one it cannot be used for the other.

I'm not so certain about that. But I don't think I have
a clear argument to the contrary, at least for now...

> 3. For management of contnent many people require a persistent identifier.

This identifier is persistent:

	http://www.w3.org/2000/10/c060d0d-103-501-b07-1090c0030e

I claim that it will never be bound to any other denotation.
True, we do not have a guarantee from ICANN of perpetual
ownership of w3.org, but we are in the process of securing
a commitment from the libraries of MIT, INRIA, and Keio
to maintain the w3.org domain as long as they are able,
If that's not persistent, then I don't know what is.



> 
> ....etc.
> 
> See the indecs model (Framework) for a lucid discussion of all these
points
> (http://www.indecs.org/results/archive.htm)

Hmm... I looked at indecs a while ago, but I didn't
see anything that refuted the idea of using URIs
for physical objects. Perhaps I'll have to take
another look...

> 
> -----Original Message-----
> From: Aaron Swartz [mailto:aswartz@upclink.com]
> Sent: 13 October 2000 22:47
> To: uri@w3.org
> Subject: URIs for Physical Items
> 
> With the increasing popularity of URIs, especially with their increased
> importance in RDF, it becomes more important than ever to have URIs for
all
> sorts of items. Already, protocols like isbn: allow the use of some types
of
> physical items, but this still leaves quite a lot left.
> 
> Would it be appropriate to use the URLs to the websites of such items? For
> example, if I wanted to talk about my iBook, pointing to
> http://apple.com/ibook/ or does this refer to the homepage, rather than
the
> item itself?
> 
> I hope this is an appropriate topic of discussion for this list. If not,
my
> apologies, and suggestion for a better place would be appreciated.
> 
> Thanks a lot for your help,
> --
> Aaron Swartz <aswartz@upclink.com>|       The Info Network
> <http://www.swartzfam.com/aaron/> |     <http://theinfo.org>
>  AIM: Jedi of Pi | ICQ: 33158237  | knowledge is power, drink it up

-- 
bind default <http://www.w3.org/People/Connolly/kb>
<mailto:connolly@w3.org> is mailbox of 
  [a Person; called "Dan Connolly";
  affiliation [ a Consortium; called "W3C";
	     homePage <http://www.w3.org> ];
  homePage <http://www.w3.org/People/Connolly/>;
  officePhone <tel:+1-913-491-0501>;
  pager <mailto:connolly.pager@w3.org> ]

Received on Friday, 27 October 2000 07:03:55 UTC